Essays in Behavioral and Household Finance

Abstract

Economists often assume that individuals are rational and that they perceive variables perfectly. Recent evidence from psychology shows that people do not perceive numerical magnitudes accurately. This dissertation studies how agents misperceive the magnitude of three key economic variables: returns of an investment, interest rates of a loan, and probabilities in risky choice. In Chapter 1, I study how investors misperceive returns of investment opportunities. When converting from dollar amounts to percentage returns and vice-versa, individuals anchor on a customary dollar amount and do not fully adjust to amounts that substantially deviate from it. I empirically test two predictions of the model in one online experiment and one field experiment with more than 170,000 investors from a brokerage firm. Having found supportive evidence for the model, I apply it to understand the effect of share prices — quoted in dollar terms — on the behavior of stock returns. Consistent with the model, I find that stocks with similar past returns, but higher share prices, exhibit stronger return predictability and are less likely to show reversal in the long-run, controlling for size. Based on this finding, I propose a trading strategy that enhances the performance of momentum strategies in the stock market. In Chapter 2, I study how borrowers misperceive the interest rates of a loan. Given a fixed interest rate, I study whether different magnitudes of the dollar amounts used to explain the interest rate affects the demand for loans. To do so, I run a field experiment with more than 130,000 clients of a Brazilian bank selling loans whose interest rates are above 100% APR. Clients are highly sensitive to the dollar amount observed, increasing the take-up of loans by 19% when observing the explanation conveyed in lower dollar amounts, as predicted by the model. In Chapter 3, I study how investors misperceive probability when making investment decisions because they anchor on their probability priors. Despite their importance in finance, these prior beliefs are usually not observable to researchers. To study the effect of these prior probabilities on perceived probabilities and risk-taking, I present a model of probability misperception in which the decision-maker anchors on a probability prior and insufficiently adjust toward the observed probability. The model generates a probability weighting function that varies according to the probability prior, making the function context dependent. I test the predictions of the model in a laboratory experiment that varies the probability prior subjects face in an investment task. In the experiment, I find that the probability prior distorts the decision-maker’s subjective probability, leading subjects to overestimate small probabilities and underestimate large ones, and that they do so to different degrees based on their probability prior. Similarly, risk-taking depends on decision-maker’s probability prior, even after accounting for the objective probability of the event. Ignoring information about differences in probability priors can lead researchers to mistakenly attribute differences in subjects' risk-taking to other observed characteristics.